Optimization of elevator group scheduling algorithm of multi-objective control

LI Qiang1， ZHENG Yong-kang2， WANG Han-tao1
（1. Institute of Computer， Hangzhou Dianzi University， Hangzhou 310018， China；
2. Hangzhou Optimax Technology Co.， Ltd.， Hangzhou 310052， China）

Abstract： Aiming at the shortcoming of lack of learning ability for fuzzy inference to obtain the index reliabilities of elevator group control，the artificial fish-swarm algorithm was led into optimize the multiple linear regression curve of the decision function. Four input variables：hall call wait time（HCWT），maximum of HCWT（maxHCWT），capacity of validation（CV）and group density（GD）could be gotten through the collection and calculation of call signals and running signals of elevators，and the three index reliabilities of elevator group control：average wait time（AWT），long-time wait percentage（LWP）and energy consumption（ENC）could be gained by fuzzy inference with these input variables. The artificial fish-swarm was trained with these three index reliabilities as samples to gain the decision function，which could be used to calculate the index reliabilities，and then be taken to linear weighted mean with the weighting coefficients which was determined by passenger traffic mode. This formula was taken as the objective evaluation function of elevator group control to reach the multi-objective control of running efficiency and energy-saving. Evaluated by the elevator group control simulation platform，the results indicate that the optimized scheduling algorithm can improve obviously the indexes of elevator group control.
Key words： elevator group control system（EGCS）； scheduling algorithm； artificial fish-swarm algorithm； fuzzy inference； multi-objective control； optimization